Interoperability among Service Registry Standards

Standards Editor : Jim Whitehead • [email protected] .edu Interoperability among Service Registry Standards Eyhab Al-Masri and Qusay H. Mahmoud • University of Guelph Basic similarities among the several service-registry standards that exist today suggest the potential for interoperability. Given that service registries are key components in Web services architectures,interoperability will play an increasingly important role in facilitating dynamic business-to-business interactions. Yet, the standards’ distinctive approaches to discovery make it unclear whether the provisional specifications will eventually merge or coexist. S ervice-oriented architectures follow the findbind-execute paradigm in which service providers register their services in public or private registries, which clients use to locate services. If it has information on a service that matches the client’s criteria, the registry provides a contract and an endpoint address. Service registries provide the foundation for service cataloging and classification; they represent a unified environment for publishing and discovering services. As Web services proliferate, registries will play an increasingly important role in connecting clients with service providers. However, the lack of efficient techniques for making Web services self-describable and easily discoverable across heterogeneous registries could undermine the driving objectives. The two main registry standards are currently UDDI (www.uddi.org) and e-Business XML (ebXML; www.ebxml.org), which is supported by the United Nations Centre for Trade Facilitation and Electronic Business and the Organization for the Advancement of Structured Information Standards (Oasis; www.oasis-open.org). Other important standards in this space include ISO/IEC 11179 (www.metadata-standards.org/11179), which serves as an international standard for specifying a metadata registry’s structure, and the Dublin Core Metadata (www.dublincore.org). These registries all contain descriptions of data or metadata, but they differ in the level of granularity and amount of syntactic and semantic information they contain. Although each has a different primary focus, 74 Published by the IEEE Computer Society these standards share basic similarities that encouraged us to explore the potential for interoperability among them. UDDI UDDI serves as a central service directory for publishing technical information about Web services; it aims to provide a platform-independent, open framework in which all businesses and their Web services meet. Initially directed by Microsoft, IBM, and Ariba, UDDI v3.0 has recently been ratified as an Oasis standard and is now endorsed by hundreds of other companies. Indeed, the virtually unprecedented level of industry cooperation in developing UDDI is helping speed the adoption of business-to-business e-commerce applications. UDDI is divided into three main categories: • White pages provide organizational contact information such as business name, address, and telephone number. • Yellow pages provide information that categorizes businesses according to standard taxonomies such as the North American Industrial Classification System and the United Nations Standard Products and Services Code. • Green pages provide technical interface definitions for Web services, including references to specifications and pointers to access points (such as URLs or email addresses). The UDDI information model contains four core elements: 1089-7801/07/$25.00 © 2007 IEEE IEEE INTERNET COMPUTING Service Registry Standards • the BusinessEntity object contains the business information, • the BusinessService object contains the service information, • the BindingTemplate object contains the binding information (how to invoke a Web service), and • the tModel object represents the service specifications. Despite numerous efforts to standardize how to publish Web services or interchange information, UDDI suffers several major shortcomings. First, it provides no guarantees regarding the quality of registered services. Moreover, registration is voluntary, which means that public registries risk becoming passive or containing outdated information. The standard’s simplified forms of search can be inadequate for articulating advanced search queries, thus forcing clients to devote extra time to finding and ranking relevant services. It also lacks business-oriented integration capabilities, such as service provider validation, life-cycle management, pricing models, semantic support, quality-ofservice metadata, security, authentication, and trust establishment. ebXML ebXML is a business-to-business XML framework designed to allow organizations to advertise and discover information about themselves. It stores Collaboration-Protocol Profile (CPP) and Collaboration-Protocol Agreement (CPA) and other information relevant to business collaboration. The CPP is an XML document that contains information about a business and the way it exchanges information with other businesses. The CPA is also an XML document that describes the specific capabilities that two businesses have agreed to use in a business collaboration. Unlike a UDDI registry, which includes only metadata, an ebXML registry can hold both metadata and arbitrary content. To illustrate, consider MAY • JUNE 2007 how the two registries publish a Web service: whereas the service provider can publish only the metadata about the service to UDDI and must maintain the actual service description on its Web site, it can publish a Web service description in an ebXML registry and repository that includes all metadata, as well as technical specifications and related artifacts. ebXML focuses on collaboration among businesses. To that end, it provides two tightly coupled components: the registry provides service interfacing, reference-system implementation and an information model, whereas the repository stores dynamic and transactional information. However, the existing ebXML standard has several limitations. Notably, it provides no guarantees regarding the quality of registered services, and it supports neither negotiation on business process definitions nor concurrent collaboration executions. Several software tools support XML-based registries. For example, the Java API for XML Registries (JAXR; http://java.sun.com/webservices/jaxr) provides a unified, standard Java API for working with UDDI and ebXML registries. In addition, several related standards have emerged to extend ebXML. For example, the Universal Business Language (UBL; www.oasis -open.org/committees/ubl) is designed to standardize common interchange documents for business-to-business integration. UBL’s main purpose is to promote standard XML schemas for basic business practices such as purchases, invoices, and shipping notices. Other Standards ISO/IEC 11179 serves as a standard for metadata registration and addresses the semantics, representation, and registration of data descriptions. It uses data-element definitions to establish meaning for specific concepts: • a data element concept (DEC) describes the object type, • a conceptual domain (CD) describes the symbolic semantics or a series of categories representing value meanings, and • a value domain (VD) defines a series of permitted values that correspond to one of the CD categories — a data element is an integral part of DEC and is associated with a particular VD. The most notable ISO/IEC 11179 implementations include the US Federal Aviation Agency’s Data Registry, the Australian Institute of Health and Welfare’s Knowledgebase, Statistics Canada’s Integrated Metadatabase, and the US Environmental Protection Agency’s MetaPro. Administered items in various application areas are registered and assigned internationally unique identifiers through registration authorities. Yet, such administrative techniques could actually hinder efforts to establish an open system in which businesses of all sizes can participate. Moreover, ISO/IEC 11179’s lack of clearly defined interfaces for accessing metadata in a registry makes it infeasible to effectively search for registered metadata. In addition, the standard currently supports symbolic data semantics but provides inadequate technical features for using metadata registries to associate the advanced technical information (regarding binding, for instance) necessary for business-tobusiness integration. Instead, it depends primarily on cross-topic XML and APIs to perform such functionality. The Dublin Core Metadata focuses on standards for interoperable online metadata; for example, the Dublin Core element set (www.dublincore.org/ documents/dces) is designed for crossdomain information description in which each element in the set is described via ISO/IEC 11179 attributes. The Dublin Core element set provides a vocabulary for defining and describing “core” information properties such as Creator, Language, Date, Title, 75 Standards and Description. The Dublin Core standard defines two levels: • Simple Dublin Core comprises fifteen metadata elements. • Qualified Dublin Core includes a group of element qualifiers that refine the meaning of a resource and three other additional elements (audience, provenance, and rightsHolder). Among Dublin Core’s many users are the National Library of Canada, US Library of Congress, US National Science Foundation, and the European Commission Diffuse Project. However, but dependence on such external sources in a multi-registry environment could potentially degrade performance and increase execution time for real-time queries because of issues related to connectivity, network latency, upgrades, authentication, and so on. Although the ability to be listed in multiple registries provides businesses the opportunity for greater exposure, it will also impact the Web services discovery process. More specifically, discovery systems will locate redundant information listed within different service registries. A possible solution is a Web services’ crawler engine that can facilitate A possible solution is a Web services’ crawler engine that can facilitate the aggregation of Web service references, resources, and description documents. the Dublin Core’s simplicity presents some serious drawbacks when working with complex or detailed collections. On the other hand, it also makes it easier to catalog information and to supplement other methods used for searching and indexing metadata. the aggregation of Web service references, resources, and description documents and provides a well-defined access pattern of usages on how to discover Web services across heterogeneous service registries such as the Web Service Crawler Engine.1 Support for Multiple Service Registries Interoperability among Service Registries Apart from the problems associated with representing and retrieving data within XML registries, decentralizing service registries and enabling organizations to operate their own (which could be deployed using different registry standards) can maximize the likelihood of a significant increase in service registries; therefore, clients will soon face the challenge of finding Web services across hundreds, if not thousands of registries. To discover Web services, systems must be able to query multiple service registries and other searchable sources, Commonalities between UDDI and ebXML registries present opportunities for interoperability between registry types. UDDI can enable the discovery of specific ebXML components, such as collaborative protocol profiles, business process schemas, and collaboration protocol agreements, and the UDDI technical committee (www.oasis-open. org/committees/uddi-spec) is currently examining ways to enable discovery of ebXML registries and framework components from within UDDI. Another way to achieve interoperability between registry standards 76 www.computer.org/internet/ might be to use a combination of characteristics from other registry types. For instance, Dublin Core registries provide RDF-based metadata registries specifically designed to supply information associated with Dublin Core element sets and other related vocabularies. To facilitate the matchmaking process between clients and providers, it would be beneficial to have a common language for aggregating information about Web services. Ideally, this should be based on authoritative metadata standards such as the XML schema definition for Dublin Core RDF metadata tags, which can be associated with specific tModels within UDDI registries. ISO/IEC 11179 focuses on defining a structure for metadata registries and fundamental attributes for specifying data-component administration. Although the standard doesn’t specify registry service bindings or provide interfacing specifications, the methods it provides for administration and retrieval of registered data also work with UDDI and ebXML. ISO/IEC 11179 metadata registries support crosssystem and cross-organizational data sharing, which can give users a common understanding of the data’s meaning and representation. In addition, service providers can potentially use the ISO/IEC 11179 classification schemes to extend those in existing registries, such as UDDI and ebXML. As Web services proliferate, ineffective discovery mechanisms make it increasingly difficult to locate them quickly and accurately. In addition, enabling Web services to interact with each other will involve actions such as comparison, composition, and orchestration, thus necessitating automation, which is possible only with Semantic Web technologies’ support. Adding semantic technologies will therefore be essential to resolve issues encountered during the automated Web services discovery process. However, partial support of semantics in current registry technologies poses the likelihood IEEE INTERNET COMPUTING Service Registry Standards that Web services will be disconnected from the Semantic Web. For example, UDDI introduces tModels to provide taxonomic representation of Web services, whereas ebXML registries can store classification hierarchies in metadata to associate semantics with registry items. Researchers on separate efforts have been working to add semantics to Web services by introducing ontology languages such as the Web Service Ontology Language (OWL-S; www.daml.org/services/owl -s) — formerly DAML-S. ISO/IEC 11179 supports symbolic semantics by using the concept of a data dictionary that defines data elements and how to compose data from simple or complex data types. Yet, the current implementation provides inadequate tools for creating data dictionaries that can be integrated with registries and other components. In addition, ISO/IEC 11179 provides no way to associate complex data structures and simple mechanisms to programmatically retrieve data elements. The ability to specify data elements, complex structures, and data types for reuse should be a fundamental feature, but many existing service registries fail to provide mechanisms for reusing custom-defined data structures and data points that aren’t contained within XML schemas. A robust registry would enable the ability to track and version such information individually and facilitate its use with other structures. he success of business-to-business integration relies on the ability to use existing service registries to provide organizations with the information they require. Yet, UDDI and ebXML provide distinctive approaches for discovery, and it remains unclear whether the provisional specifications will merge or coexist, especially given their limitations in terms of the discovery of businesses and Web services. ebXML extends the discovery process by introducing e-business col- T MAY • JUNE 2007 laboration, which will be a key element in e-business. ISO/IEC 11179 provides a way to register individual data elements, ebXML allows for registering XML schema, and UDDI allows for registering Web services. Ideally, clients should be able to efficiently query and extract components such as data elements, XML schemas, and Web services from within the same registry or a group of registries. One promising approach is to enhance the discovery process to work with ebXML framework components from within UDDI and vice versa, using ISO/IEC 11179 classification schemes to augment the publication of Web services, and incorporating the Dublin Core metadata tags to improve the matchmaking process within service registries. We are currently working to enhance Web services discovery across heterogeneous registries, such as the Web Service Repository Builder.2 Our proposed framework is compatible with and can be integrated seamlessly into existing service registry infrastructures without any modifications to existing environments. Interoperability among registries would complement the strengths of each standard, although the ability to administer, manage, and discover services in a unified fashion across heterogeneous service registries remains an obstacle as services proliferate. References 1. E. Al-Masri and Q.H. Mahmoud, “A Framework for Efficient Discovery of Web Services across Heterogeneous Registries,” Proc. IEEE Consumer Communications and Networking Conf. (CCNC), IEEE CS Press, Jan. 2007. 2. E. Al-Masri and Q.H. Mahmoud, “Crawling Multiple UDDI Business Registries," Proc. 16th Int'l World Wide Web Conf., ACM Press, May 2007; http://www2007.org/program/ poster.php?id=968. Eyhab Al-Masri is a PhD student in the Department of Computing and Information Science at the University of Guelph. His research interests include service-oriented computing and Web services. He has a BS in computer engineering and an MS in electrical engineering, both from Florida International University. Contact him at ealmasri@ uoguelph.ca. Qusay H. Mahmoud is an associate professor in the Department of Computing and Information Science at the University of Guelph, and associate chair of the Distributed Computing and Wireless & Telecommunications Technology program at the University of Guelph-Humber. His research interests include service-oriented computing, agent technology, middleware, and mobile computing. Mahmoud has a PhD in computer science from Middlesex University. Contact him at [email protected]. SEE THE FUTURE OF COMPUTING NOW in IEEEIntelligent Systems Breakthroughs in areas such as intelligent agents, the Semantic Web, data mining, and natural language processing will revolutionize your work and leisure activities. Read about this research as it happens in IEEE Intelligent Systems. www.computer.org/ intelligent/subscribe.htm 77